Back to News
InnovationAI Understanding briefing

Underdog Unveils Woof, a 32MB AI Speech Model for Browser Use

Underdog has released Woof, a 32MB AI speech model that runs in web browsers. The model has about 31 million parameters and is available as open weights under a non-commercial license. It can run on most devices without an internet connection.

4 min readRead the linked source
Source-provided image accompanying Underdog Unveils Woof, a 32MB AI Speech Model for Browser Use
Source referenceSource recorded
Publisher
adgully.com
Source type
Linked source — primary-source status has not been established.
ContextUnderstand this in 60 seconds

What happened

Underdog has released Woof, a 32MB AI speech model that runs in web browsers. The model has about 31 million parameters and is available as open weights under a non-commercial license. It can run on most devices without an internet connection and has been demonstrated through a browser game.

Underdog has released Woof, a 32MB AI speech model that runs in web browsers.

The model has about 31 million parameters and is available as open weights under a non-commercial license.

It can run on most devices without an internet connection and has been demonstrated through a browser game.

The game is set in a retro room with a typewriter, a CRT monitor, and a desk lamp.

Players type text on the typewriter and hear it spoken, and a tuner generates different kinds of music.

Source details: adgully.com

Why it matters

The release of Woof is significant because it provides a small and efficient AI speech model that can run on users' own devices. This aligns with Underdog's pitch of AI that runs on users' own devices, rather than relying on cloud providers. The model's performance has not been independently verified, but Underdog claims it beats models 1,000 times its size on word error rate. The release of Woof follows the company's previous release of Bark, a speech-to-text model of about 165 MB for transcription and dictation. Underdog has also released a 27-billion- model that runs on laptops, along with a waitlisted app positioned as a locally hosted alternative to Muse and Instinct.

The release of Woof is significant because it provides a small and efficient AI speech model that can run on users' own devices.

This aligns with Underdog's pitch of AI that runs on users' own devices, rather than relying on cloud providers.

The model's performance has not been independently verified, but Underdog claims it beats models 1,000 times its size on word error rate.

The release of Woof follows the company's previous release of Bark, a speech-to-text model of about 165 MB for transcription and dictation.

Underdog has also released a 27-billion- model that runs on laptops, along with a waitlisted app positioned as a locally hosted alternative to Muse and Instinct.

Interactive Mechanism

Interactive Mechanism: How It Actually Works

Explore the underlying technology behind this development interactively.

Model Parameter Size:8B Parameters
VRAM Required5.5 GBGPU memory footprint
Target HardwareMacBook / Single GPUDeployment tier
Privacy100% Air-GappedLocal device capability
Core takeaway: Small, quantized models (3B–8B) now run directly inside smartphones and laptops with complete data privacy, while mammoth 400B+ models remain the domain of datacenter clusters.
Interactive Concept Check+10 Points
What is AI? Quiz

A route planner searches possible journeys using explicit rules. What does this illustrate about AI?

What to watch next

The impact of Woof on the AI industry and its potential applications will be worth watching. The model's performance and usability will also be important factors to consider. Underdog's goal of training the best American-made open models that people can run privately on devices they already own is an interesting development. The company's pitch of AI that runs on users' own devices, rather than relying on cloud providers, is also worth noting.

The impact of Woof on the AI industry and its potential applications will be worth watching.

The model's performance and usability will also be important factors to consider.

Underdog's goal of training the best American-made open models that people can run privately on devices they already own is an interesting development.

The company's pitch of AI that runs on users' own devices, rather than relying on cloud providers, is also worth noting.

The release of Woof and other models by Underdog will be worth monitoring for further developments.

Related guides & quizzes

Found this useful?